Syracuse University
SURFACE
Syracuse University Honors Program Capstone
Projects
Syracuse University Honors Program Capstone
Projects
Spring 5-1-2011
Investigating How Resource and Situation Type
Influence the Sunk-Cost Fallacy
Jenae A. Richardson
Follow this and additional works at: http://surface.syr.edu/honors_capstone
Part of the Other English Language and Literature Commons, and the Psychology Commons
Recommended Citation
Richardson, Jenae A., "Investigating How Resource and Situation Type Influence the Sunk-Cost Fallacy" (2011). Syracuse University
Honors Program Capstone Projects. Paper 296.
This Honors Capstone Project is brought to you for free and open access by the Syracuse University Honors Program Capstone Projects at SURFACE.
It has been accepted for inclusion in Syracuse University Honors Program Capstone Projects by an authorized administrator of SURFACE. For more
information, please contact [email protected].
1
The Sunk-Cost Fallacy
Imagine that you just rented a DVD from Redbox at Target. You drove
through traffic for an hour and arrived at the store for the DVD only to find
out that you have to wait 15 minutes for the machine to be fixed because it is
broken. You travel home, stick the DVD into its player, and begin to watch
the movie. After 10 minutes, you realize you’re bored to tears. Do you stop
watching the movie immediately, wait to see if it gets better, or watch it until
the end?
If you continue watching the movie—even though you are bored—
you’ve committed the sunk-cost fallacy. Derived from economic theory and
with applications in psychology, the sunk-cost fallacy refers to a decision
making bias that people often have to continue investing future resources,
whether time, money, or effort, into something they have already made prior
investments in (Arkes & Blumer, 1985). The sunk-cost effect has been
studied mainly regarding hypothetical decisions of everyday problems (Frisch
1993; Klaczynski, 2001; Stanovich, 1999), for example like the problem
stated above: Continuing to watch a boring movie until it ends because you
paid for it (Strough et al., 2008). The sunk-cost effect has also been studied in
real-life settings. Staw and Hoang (1995) investigated how the sunk-cost
effect influenced how much NBA players were paid and how long they
remained hired despite performance. The researchers found that despite
players’ on-court performance or number of injuries, highly drafted players
were retained longer. Because the sunk-cost effect is shown in both
2
hypothetical and real-life situations, hypothetical situations are valuable and
efficient tools for examining variables that might affect the sunk-cost fallacy.
Although an abundance of sunk-cost research has focused on
individual situations, or situations in which a person’s decision is not the
result of his or her relationship with someone else (such as deciding to watch a
boring movie or buying season tickets for a game), there is scant literature
about how the sunk-cost effect may influence interpersonal situations.
Coleman (2009) studied how the sunk-cost effect was related to commitment
to dates arranged online. He found that as prior investments increased, the
participants’ likelihood to commit to a blind date increased, even if the date
was described as “inferior.” In a similar study, the sunk-cost fallacy was
examined in real-life relationships. Rusbult (1980a, 1983) found that the
larger intrinsic investments (money, effort, time, emotions) were in the current
relationship, the less likely a participant would be to express interest in dating
an alternative person. Even in cases in which participants reported high
relationship costs, however, participants still reported being committed to
their relationships if prior investments had been made.
Individual v. Interpersonal Situations
As previous research illustrates, the sunk-cost effect has implications
for decision-making behavior in both individual and interpersonal situations.
To date, no study has yet investigated how resources invested into both
situation types might differ because of the situation. For example, could an
investment of money drive the sunk-cost effect in an individual situation but
3
have no effect in an interpersonal situation? This type of question has yet to be
answered by sunk-cost research and was a goal of the present study.
Perhaps one major difficulty in answering this question results from an
unclear understanding of how resource types influence the sunk-cost effect.
Researchers have noted that time, money, and effort predict the sunk-cost
effect, but most researchers have not shown how each of these resources
individually influences the sunk-cost effect (Arkes & Blumer, 1985). One
article, however, has shown how investments of time and money could have
different influences on the sunk-cost effect (Soman, 2001). Soman found that
participants were more likely to commit the sunk-cost fallacy for prior
investments of money than for prior investments of time, unless time was
expressed in the form of monetary quantities. In addition, Coleman (2010a)
extended Soman’s research by investigating how an investment of effort was
similar to and different from time and money investments. He found that prior
investments of money predicted the sunk-cost fallacy, but prior investments of
time and effort did not.
While Soman and Coleman’s research aids sunk-cost research by
illustrating how different resources affect sunk costs, more research is needed
to investigate additional resource types and their effects on sunk costs.
Research by Goodfriend and Agnew (2008) suggests that emotions and selfconcept may also be resources invested into a romantic relationship and that
these resources may be predictive of the sunk-cost effect. While emotions and
self-concept are explored as investments in the research that links romantic
4
relationships with the sunk-cost effect, these investments are nearly absent
from research on the sunk-cost effect in individual situations. To extend
Coleman’s research, an investigation of these two investment types in addition
to the most commonly cited three (time, money, effort) could broaden
peoples’ understanding of the sunk-cost fallacy in both types of situations.
Investigating Another Type of Interpersonal Relationship
In addition to the paucity of articles describing how investment types
directly influence the sunk-cost effect, another major limitation is the lack of
research about how the sunk-cost effect affects friendships. Most of the
research done on interpersonal relationships has focused on romantic
relationships. There is some research that suggests, however, that the sunkcost effect may also influence friendships.
Rusbult (1980b) studied commitment along with satisfaction and
alternatives in college friendships. She found that prior investments helped to
predict future commitment in friendships. Rusbult’s research suggests that like
romantic relationships, friendships may also be vulnerable to the sunk-cost
effect if people have invested resources such as time, effort, and emotions that
are inextricably connected to the friendship and thus are hard to recover
(Rusbult, 1980b). Examining both types of interpersonal relationships in the
same study could be helpful in determining exactly which individual resources
may predict future commitment in both relationship types.
5
Psychological Motives Underpinning the Sunk-Cost Effect
Researchers posit that the sunk-cost effect results because of the
“Don’t waste” rule or the idea that people are unable to ignore the costs that
have already been sunk into an endeavor. Thus, they continue to invest more
resources—or escalate commitment in response to sunk costs (Arkes &
Ayton, 1999; Staw, 1981). In one study, Arkes and Blumer showed how the
“Don’t waste rule” accounted for decisions made. The team asked participants
to imagine spending $100 on a ticket for a weekend ski trip to Michigan and
then several weeks later buying a $50 ticket for a weekend ski trip to
Wisconsin. Participants were specifically told: “You imagine enjoying the
Wisconsin trip more than the Michigan trip, but then you notice that each trip
will occur on the same weekend and that it’s too late to sell or return either
ticket.” Participants were asked to select which trip they would go on. The
researchers found that participants selected to go on the less desirable
Michigan trip, because they thought going on the Wisconsin trip albeit less
expensive would “waste” twice as much money (Arkes & Blumer, 1985).
Researchers continue to investigate the psychological motivations
underpinning escalating commitment to a course of action, or continuing to
work on an assignment even when the costs are evident or the payoffs aren’t.
Economist Glen Whyte (1986) found that escalating commitment could be
explained through prospect theory or the idea that people frame decisions
from a reference point and the decision to escalate commitment is based on
both negative and positive consequences. In his model, Whyte shows that if an
6
action has negative consequences, a person’s decision to commit further
resources is framed as a choice between losses, whereas if the action has
positive consequences, the choice to continue investing further resources is
framed as a choice between both gains. Despite either consequence, Whyte
posits that either consequence would lead to escalation commitment because
negative consequences would lead to risk-seeking behavior and positive
consequences to risk-averse behavior (Tversky & Kahneman, 1981; Whyte,
1986).
Escalating commitment is directly linked to the sunk-cost effect,
because prior investments usually enter into future decisions even in the face
of a course of action that’s going terribly awry (Staw, 1981). Economist Barry
Staw (1981) explained that people might escalate commitment because of
self-justification or the need to justify prior choices (Brockner, 1992; Staw,
1976; Staw, 1981). Self-justification is directly linked to the sunk-cost effect,
because in the need to justify prior choices, people often invest additional
resources. An illustrative example that Staw describes in his study is an
administrator that has allocated research funds to an operating division of a
company and tries to justify his or her potentially ineffective decision by
escalating commitment and committing future resources to the project, thus
committing the sunk-cost fallacy.
Predictions of Five Resources
The present study investigated which resources: time, money, effort,
emotion, or self-concept, would be most likely to lead people to commit the
7
sunk-cost fallacy in an individual and interpersonal situation. Because there is
limited research about how each resource type influences the sunk-cost effect,
I extrapolated from other areas and concepts in psychology in order to make
predictions about the role each resource would play in influencing the sunkcost effect.
Money
An entire body of sunk-cost literature supports the prediction that
money leads people to commit the sunk-cost fallacy in individual situations
(Arkes & Blumer, 1985; Phillips, Battalio & Kogut, 1991; Staw, 1981).
Soman (2001) suggested that this might be the case, because people are better
at mentally accounting for money than they are for time. Some of the reasons
he listed for a difference in the sunk-cost effect for money and time were: time
cannot be inventoried or replaced; time is not as easily aggregated as money;
and accounting for money (unlike accounting for time) is a routine activity
(Soman, 2001). In contrast, Goodfriend and Agnew (2008) found that money
was not a predictor of commitment in romantic relationships, because couples
often value nonmaterial resources over material resources such as money and
because happiness is more directly linked to nonmaterial resources such as
emotions and time (Diener, et al., 1999; Goodfriend & Agnew, 2008). This
research helped me to predict that money would increase the likelihood of
people committing the sunk-cost fallacy in individual situations but wouldn’t
increase the likelihood of people committing the sunk-cost fallacy in
interpersonal situations.
8
Time
Soman (2001) predicted that time wouldn’t lead participants to commit
the sunk-cost fallacy, because individuals account for time and money
differently. In six separate experiments, he gave participants scenarios in
which the only thing he varied was a time or money investment. Soman found
that while participants committed the sunk-cost fallacy for money, the sunkcost effect disappeared for time investments. Soman concluded that
participants might be unable or unwilling to account for time in the same way
that they account for money (Soman, 2001). While Soman’s research provides
evidence that time does not lead to the sunk-cost effect in individual
situations, research on interpersonal relationships suggests that time is an
intangible investment that predicts the sunk-cost effect (Goodfriend & Agnew,
2008). Thus, based on this research, it was logical to predict that time would
not lead people to commit the sunk-cost effect in the individual situation used
in this study, but that time would lead to the sunk-cost effect in the
interpersonal situation.
Emotion
Little research has been done to explain what effect invested emotions
might have on the sunk-cost fallacy. Coleman (2010b) investigated the role
emotions would play on participants’ likelihood to continue signing up online
for a class. He found that participants that were induced to feel anger
committed the sunk-cost effect but that participants that were induced to feel
9
fear did not commit the sunk-cost fallacy. Coleman reasoned that the findings
might have occurred because participants that are angry are likely to feel that
they have control of the situation, whereas those that are fearful might have
less circumstantial control. As a result, those who perceive themselves as
having more control might be more likely to escalate commitment, thus
committing the sunk-cost fallacy, while those who feel that they have little
control are less likely to escalate commitment (Coleman, 2010b).
Although Coleman’s research might be helpful in predicting how
induced emotions might predict the sunk-cost effect, his research does little to
suggest how emotions invested in interpersonal relationships could predict the
sunk-cost effect in that context. Some research suggests that emotions may
lead people to remain in relationships that they aren’t necessarily satisfied
with (Rusbult, 1980a), although Rusbult doesn’t explicitly state why this
might be the case. Other research supports the idea that emotions exert an
important influence on cognition and decision-making across everyday
problems (Bower, 1981; Labovie-Vief, 1992; Sinnott, 1989).
Effort
Coleman (2010a) investigated whether effort would lead participants
to commit the sunk-cost fallacy. Coleman found that invested effort did not
produce the sunk-cost effect. Coleman reasoned that people aren’t good at
mentally accounting for effort because they’re not well versed in keeping
track of this type of investment. Coleman also explained that there were
alternative explanations for the sunk-cost effect other than the sunk-cost effect
10
such as cognitive dissonance, self-perception theory, and learned
industriousness. Coleman’s research provides support for my prediction that
invested effort will not lead people to commit the sunk-cost fallacy for
individual situations. Effort, however, is described as an intangible investment
that leads people to commit the sunk-fallacy in interpersonal relationships
(Goodfriend & Agnew, 2008). As a result, I predicted that effort would lead to
the sunk-cost effect in interpersonal relationships in our study.
Self-Concept
Perhaps the most under investigated area of sunk-cost research deals
with how the self-concept is related to the sunk-cost fallacy. We were
particularly interested in investigating self-concept, most simplistically
defined as “a person’s perception of him- or herself. Self-concept is
influenced by peoples’ interactions with their environments and by
attributions for their own behavior (Marsh & Shavelson, 1985; Shavelson,
Hubner, & Stanton, 1976).
Peoples’ self-perceptions could potentially influence whether they
escalate commitment to a particular course of action. Coleman suggested this
when he argued that there were alternative explanations of effort other than
the sunk-cost effect (Coleman, 2010a). One of the theories he discussed—
cognitive dissonance theory—could provide the link between the sunk-cost
fallacy and self-concept. Some research on cognitive dissonance describes it
as a psychological discomfort that people are motivated to decrease (Elliot &
Devine, 1994). Oftentimes, cognitive dissonance results when an individual’s
11
actions conflict with his or her view of his or herself and must resolve that
conflict by changing his or her behavior or adding a cognition that’s consistent
with one’s view of his or herself (Festinger, 1957).
If participants view themselves as caring and supportive and are told
that a friend or romantic partner has stopped talking to them and participants
are asked to select how much longer they’d remain in the relationship,
cognitive dissonance theory implies that participants may choose a behavior in
line with their self-perceptions. In this particular example, participants may be
more likely to remain in the relationship in order to reduce any dissonant view
of themselves as distant. This escalation of commitment suggests that people
with this type of self-perception are likely to commit the sunk-cost fallacy. On
the flip side, people that view themselves as distant and detached in a
friendship or relationship would be unlikely to change their behavior during
cognitive dissonance. Thus, we can expect that someone who views his or
herself that way would end the relationship, thus not committing the sunk-cost
fallacy. All in all, I predicted that participants in the high investment condition
for self-concept (caring and supportive) would commit the sunk-cost effect,
while participants in the low invest condition would not fall prey to the sunkcost effect.
Personality Variables
We were interested in investigating how personality could moderate
the sunk-cost fallacy. Research about personality traits and the sunk-cost
effect were limited, so we looked at articles that linked personality traits to
12
escalation of commitment. Moon et al. (2003) divided neuroticism into
anxiety and depression and investigated how those facets of the neuroticism
construct influenced escalation of commitment. The authors found a
moderately significant effect of neuroticism on escalation of commitment.
Anxious individuals were more likely to escalate commitment while
depressed individuals were not. Conscientiousness, however, did not predict
escalation of commitment, although achievement strivers were more likely to
escalate commitment (Moon, 2001). Based on these studies, we predicted that
neuroticism and conscientiousness would predict the sunk-cost effect, because
they may lead participants to escalate commitment.
The Present Study
The present study sought to address the limitations of extant sunk-cost
research. A major goal of the present study was to answer the question:
Which type of invested resources would result in the highest sunk-cost fallacy
scores? To do so, I investigated five resources that people likely invest in
individual and interpersonal situations. The current study also sought to
investigate how these resources may be alike or different in individual and
interpersonal situations, so these situations were investigated simultaneously
for comparison purposes. Another goal of the present study was to identify
whether any personality traits predicted the sunk-cost effect.
There are two independent variables in this study. One independent
variable is the situation type (or context) that was divided into two levels:
individual and interpersonal. These levels were further divided into romantic
13
relationship and friend, two important types of interpersonal relationships.
Another independent variable is the type of invested resources that we divided
into five levels: time, money, effort, self-concept, and emotions. There are two
dependent variables. One dependent variable is the sunk-cost fallacy. The
second dependent variable is personality that is measured using a 10-item
inventory.
The present study sought to accomplish its goals by investigating how
the sunk-cost effect influences a population of college students, an age group
that may be particularly vulnerable to the sunk-cost effect based on prior
research that found that younger adults are more likely than older adults to
commit the sunk-cost fallacy (Strough et al., 2008). The present study is also a
pioneer in determining what types of investments are important to collegeaged students—both in individual and interpersonal situations.
Method
Design
A 3x5 mixed-factorial design was used. The between-subjects variable
was resource. Participants were randomly assigned to receive one of the five
resources (money, time, emotion, effort, self-concept). The within-subjects
variable was situation. Each participant answered questions based on 3 levels
(exam, romantic relationship, and friend). The dependent variable was the
sunk-cost fallacy score.
14
Participants
Students at Syracuse University (N=207) were recruited through an
email announcement, via SONA, and by word of mouth. Participants were
brought into the laboratory where they completed the study. They volunteered
or received research credit or extra credit for their psychology classes as
compensation for participating. There were 81 male and 126 female students
that ranged in age from 18 to 25 (M= 19.54, SD= 1.39).
Procedure
Participants were brought into the laboratory and were seated at a table
where they completed the informed consent form. After participants
completed the consent form, I selected a slip of paper numbered Version 1
through 30. Participants were randomly assigned to complete a version, each
of which was counterbalanced. Sunk-cost fallacy scores did not significantly
differ by version. There were five versions for each resource. Each participant
completed a series of questions for three situations and based on one resource.
Each situation had two questions that were identical except that one involved
high investment of a resource and the other, a low investment. In all, each
student completed six questions for Packet A.
The individual situation asked them how much longer they would
continue to take a standardized exam if they hadn’t gotten the score they
needed the first time around. The friendship and romantic relationship
vignettes were modeled similarly. Participants were asked how much longer
15
they would remain in a friendship or romantic relationship in which their
friend or partner had recently stopped responding to text messages, phone
calls, and was not supportive of their goals and interests.
Next, participants completed a 10-item personality inventory. Finally,
participants completed a demographics questionnaire that asked them to report
such things as their household income and whether or not they had been in a
romantic relationship or friendship in which their partner or friend stopped
talking to them. In addition, participants were asked to rate the five resources
in order of importance for each situation type. After participants completed
the questionnaire, their data was entered into a computer, and the sunk-cost
fallacy was calculated and personality inventories were scored.
Measures
Decision vignettes. Vignettes were similar to the ones used by Strough
et al. (2008) that they adopted from Frisch (1993). The scenarios included
different situation types such as taking a standardized exam after failing to
meet a desired goal the first time and remaining committed to a friendship or
romantic relationship that was going bad. An example of the vignette for the
individual situation for effort was: “You spent a lot of effort preparing for a
standardized exam. After taking the exam, you received a score lower than
what’s required to be admitted to your program of choice. Considering all the
factors involved, how many more times will you take the exam?” There were
five choices for participants to select from: don’t take it again, take it one
more time, take it two more times, take it three more times, or take it
16
indefinitely (until I get the score I need). Participants received a low
investment and high investment version of each situation (See Appendix). An
exam situation was selected because it is a situation in which college students
have experience and are likely to be able to relate to.
Vignettes for the interpersonal situations were similar, and the only
words that were changed in each were friend and partner. An example of an
interpersonal situation vignette for money was: “You’ve been friends with
someone and have spent about $500 on gifts for your friend. Lately, your
friend hasn’t been returning your calls or texts messages and has not been
supportive of your goals and interests. How much longer will you remain
friends?” Participants could select from five choices: end the friendship
immediately, wait for a couple weeks to see if the friendship improves, wait
for a month to see if the friendship improves, wait for 6 months to see if the
friendship improves, or remain committed to the friendship. Participants also
received a low investment and high investment question of each situation. In
this particular example, the low investment condition for money was $40.
Computing the sunk-cost fallacy. To compute the sunk-cost fallacy
score, each participant’s decisions for the low investment and high investment
questions were compared. If a participant indicated that he or she would spend
more time for the high investment than for the low investment choice, a score
of 1 was assigned to indicate that the sunk-cost fallacy occurred. In addition,
we computed normatively-correct decision scores (Klaczynski, 2001). The
normatively correct decision was made if a participant chose the same answer
17
choice for the low and high investment conditions. A score of 0 was assigned.
Finally, an error score was computed. Participants made errors if they
indicated they would spend more future time in the low investment condition
than in the high investment condition. Only sunk-cost fallacy scores are
analyzed in this paper.
Ten-item personality inventory. We included a 10-item personality
inventory known as the Big Five Inventory-10 (BFI-10) (Rammstedt & John,
2007). The BFI-44 is shown to have high convergent validity with the
established measures of the Big Five personality traits (John, Naumann, &
Soto, 2008). The BFI-10 is a reliable and valid alternative to the BFI-44 for a
brief personality assessment (Rammstedt & John, 2007). There are 10 phrases
on the inventory. Participants were asked the question, “How well do the
following statements describe your personality?” Some of the phrases that
participants selected from were: “…is reserved,” “…is generally trusting,”
“tends to be lazy…” Participants rated their responses on a 5-point Likert-type
scale. An answer of 1 indicated that participants disagreed strongly. An
answer of 5 meant that participants agreed strongly.
Results
Resource Type and Situation Type Interaction
A mixed 3x5 ANOVA was used for the between-subjects independent
variable, situation type with three levels (individual, romantic relationship,
and friendship), and the within-subjects independent variable, resource type
18
with five levels (money, time, emotion, effort, and self-concept) and to
determine both of the independent variables’ effect on the dependent variable,
sunk cost fallacy scores.
Significant main effects of resource type F(4,200) = 8.12, p < .001,
ηp2 = 0.40 and situation type F(2,400) = 16.54, p <.001, ηp2 = .076, were
qualified by a significant two-way interaction of Resource X Situation
F(8,400) = 5.69, p < .001, ηp2 = .140. In order to localize the interaction
effect, post-hoc tests were performed to assess the simple effect of situation at
each level of resource. A Bonferroni-type correction was used to reduce Type
I errors associated with multiple post-hoc tests; alpha was set at .05. To follow
up significant simple effects of situation, I performed multiple comparisons of
situation at each level of the significant simple effect of resource type.
Money
The simple effect of money was statistically significant F(2,82) = 3.69,
p < .05. There was no statistically significant difference (p = 1.00) between
the means for the exam situation (M = .23, SD= .421) and the friend situation
(M = .44, SD =.497), but there was a marginal difference between the exam
and romantic relationship situations (M = .43, SD = .496). The difference in
sunk-cost fallacy scores between the interpersonal relationships was nonsignificant (p = .17).
19
Time
The simple effect of time was statistically significant F(2,80) = 37.99,
p < .001. Students that received the time resource (p < .001) were more likely
to commit the sunk-cost fallacy for the romantic relationship situation (M =
.43, SD = .496) and friend situation (M = .44, SD =.497) than for the exam
situation (M = .23, SD = .421). The difference in sunk-cost fallacy scores
between the two types of interpersonal relationships was non-significant (p =
1.00).
Emotion
The simple effect of emotion was non-significant F(2,80) = 2.53, p =
.086. The means of participants’ sunk-cost fallacy scores did not significantly
differ among the romantic relationship situation (M= .43, SD= .496), friend
situation (M= .44, SD=.497), or exam situation (M = .23, SD= .421).
Effort
The simple effect of effort was non-significant F(2,80) = 1.58, p =
.212 The means of students’ sunk-cost fallacy scores did not differ
significantly among the romantic relationship situation (M= .43, SD= .496),
friend situation (M= .44, SD=.497), or exam situation (M = .23, SD= .421).
Self-Concept
The simple effect of self-concept was non-significant F(2,78) = .275, p
= .760 . The means of students’ sunk-cost fallacy scores in the romantic
20
relationship situation (M= .43, SD= .496), friend situation (M = .44, SD
=.497), and exam situation (M = .23, SD = .421) were not significantly
different from each other.
Personality
Preliminary multiple linear regression analyses showed that
personality traits did not predict sunk-cost fallacy scores. I also examined
neuroticism and conscientiousness as covariates in our analysis of variance.
They were non-significant.
I conducted an exploratory analysis to find out whether sunk-cost
scores differed by age or sex. Although age was not a significant factor, sunkcost fallacy scores did differ by a Sex x Situation interaction, F(2,390) = 3.58,
p < .05, ηp2 = .02. Follow-ups showed that men had higher sunk-cost fallacy
scores for the exam situation, women had higher sunk-cost scores for the
friend situation, and there was no sex difference for the romantic relationship
situation. Because there was no main effect of sex and sex did not interact
with resource, nor was there a three-way interaction of sex, situation, and
resource, I did not include sex as an additional factor in our main analyses.
Discussion
The present study investigated the relationship between situation and
resource type and hypothesized that there would be a significant interaction
between the two independent variables. My hypothesis was supported,
revealing that the simple main effect of situation type changed over the five
21
levels of resource or that the simple main effect of resource type changed over
the three levels of situation. To determine more information about the
interaction, I conducted further analyses. A closer examination revealed
results that we I did not necessarily predict.
Money
Students differed in their likelihood to commit future resources to the
situations in the present study when previously invested money was involved.
In particular, students tended to be more likely to invest further in a romantic
relationship when money had been spent on a partner than to invest further in
taking an exam after money had been spent. The finding that money could
predict the sunk-cost effect in a romantic relationship is inconsistent with
research by Goodfriend and Agnew (2008) who argued that money is a
material resource that is not linked to happiness in relationships.
Prior research about the effects of resources on the likelihood to
escalate commitment in interpersonal relationships has mainly been anecdotal
(Lala, 2005). It might be possible that while college students report that
money does not—or should not—predict future commitment in a romantic
relationship, the reality is that money might predict future commitment in a
population of college students. This result might have been easier to explain if
we knew how participants viewed money. Perhaps, students might have
viewed money as an intangible investment in a romantic relationship—
something that they could not recover. Thus, they might have been more
likely to escalate commitment in the relationship, because as Goodfriend and
22
Agnew (2008) found, intangible investments are more likely to predict future
commitment in a romantic relationship.
The finding that students commit the sunk-cost fallacy (albeit to a low
degree) for the exam situation is supported by previous research that has
found that prior investments of money predict the sunk-cost effect (Arkes &
Blumer, 1985). However, I predicted that students would commit the sunkcost fallacy more for the exam (or individual) situation than for the
interpersonal situations. This prediction was not supported in the present
study.
There are several reasons why sunk-cost fallacy scores for the resource
of money were lowest in the exam situation. It is important to point out that if
the exam situation was not compared against the interpersonal situations, then
I might have concluded that money predicted the sunk-cost effect in an
individual situation. Aware of this limitation of prior research, I wanted to
compare an individual situation against interpersonal situations to observe
differences—or similarities between situation types. Perhaps a prior
investment of money did not really motivate students to continue investing
future resources to a failing course of action (an exam that they performed
poorly on).
Although previous research has not yet investigated this possibility,
students perhaps felt that their chances of future gain (doing well if they
continued taking the exam) were bleak. Staw (1981) explained that people
might reduce future commitment to a course of action if they perceive that
23
future gains are nearly impossible. Thus, they may be less likely to commit
the sunk-cost fallacy. The type of individual situation we used in this study
may also help to explain our results. We constructed an exam vignette,
because about all college students have taken standardized exams and many
college students will likely take standardized exams to obtain admittance to a
graduate school, law school, or medical school. Metalsky et. al (1982) studied
how attributional styles affected students’ reactions to a low exam grade. The
researchers found that students with an internal and global attributional style
(likely to attribute their poor exam grade to their performance) were more
likely to report being depressed than were students with an external and
specific attributional style (likely to attribute their poor exam grade to a bad
day or a poorly written test, for example). Results from this study suggest that
attributional styles along with invested money may predict students’ future
commitment to taking an exam. Future studies should investigate this
relationship.
Why other researchers suggest that money predicts the sunk-cost effect
in all individual situations requires further investigation. As our results
suggest, students may have perceived that doing better was beyond their
control even though money had been invested. Perhaps in other sunk-cost
situations in the extant literature, students felt greater personal control over
their course of action or maybe didn’t care. For example, in Coleman’s study
of college students’ commitment to medical treatment, he asked students to
imagine going to a chiropractor’s website and finding out that a special deal
24
was being offered but later finding out that there’s a physical therapist that has
a better chance of curing their condition (Coleman, 2010a).
While Coleman’s study is ambitious, it might be possible that college
students might find it harder to imagine or personally identify with a situation
that many of them have never been placed in, while they might be more likely
to identify with an exam situation. Thus, Coleman’s finding that money
predicted the sunk-cost effect for the medical treatment situation may be the
result of students imagining having personal control over a situation that they
did not particular identify with, thus students were more likely to commit the
sunk-cost fallacy. Future research should compare two situations in which
students feel they have personal control or lack control over the situation and
see how these differences may predict the sunk-cost effect.
Overall, the finding that college students slightly differed in how they
committed the sunk-cost fallacy for the exam situation and romantic
relationship situations suggests that college students might mentally account
for money differently between both situation types. Future research should
investigate this possibility.
Time
The finding that students were significantly more likely to remain
committed to an interpersonal relationship even when a friend or romantic
partner had stopped communicating with and supporting them is important.
This finding is consistent with research by Goodfriend and Agnew (2008) that
found that time is an intangible investment that may predict future
25
commitment in romantic relationships.
Although participants did commit the sunk-cost fallacy for time in the
exam situation, these participants’ mean sunk-cost fallacy scores were the
lowest for any of the resources, suggesting that time might lead to a lesser
tendency of people escalating commitment to a previous course of action in an
individual situation. This finding can be supported by recent research by
Soman and Coleman who found that people were less likely to commit the
sunk-cost fallacy for time, because people can’t mentally account for time in
the same way that they can account for money (Soman, 2001; Coleman,
2010).
Soman explained that investments of time did not motivate people to
remain committed to a failing course of action, unless a particular situation
was expressed in monetary terms. It is possible that mental accounting played
a role in our study, whereby participants were unable to adequately mentally
account for time. This could explain why participants with the time resource
for the exam situation had the lowest sunk-cost fallacy scores. However,
because we did observe a sunk-cost effect for time for the exam situation, that
might suggest that maybe participants could be better at mentally accounting
for time in situations that are particularly relevant to them. For example, while
it may be difficult for participants to imagine escalating commitment to
inferior medical treatment in Coleman’s hypothetical situation, they might be
better able to account for time for a situation in which most of them have
probably experienced—studying for an exam and doing poorly and
26
considering whether or not to retake it.
Our findings regarding how investments of time could differ between
situation types further illustrates the purpose of the present study, which was
to show that resources differ depending on context. Thus, researchers that
have studied the sunk-time cost effects in individual contexts should perhaps
make a caveat based on our findings: People are better at mentally accounting
for time in interpersonal relationships, although they aren’t as well-versed in
mentally accounting for time in individual situations. Future research should
compare an individual and interpersonal situation that both involve explicit
mental accounting and investigate whether there are differences or similarities
in participants’ future commitment.
Emotion
Students were just as likely to commit the sunk-cost fallacy for the
exam situation as they were for the interpersonal situations. This suggests that
emotions about the exam likely drove participants to commit future resources
to taking the exam repeatedly. To date, the only study that investigates the
role of emotion on the sunk-cost effect in an individual situation is Coleman’s,
who found that participants who were made angry were more likely to commit
the sunk-cost fallacy (Coleman, 2010).
Although we are limited in interpreting how students felt as they
imagined having to make the choice of taking an exam again after doing
poorly, it’s possible that as students placed themselves in the situation, they
became angry with themselves for having done poorly the first time and as a
27
result, were more likely to keep investing future resources to taking the test.
Students who felt upset about receiving an inadequate exam score were
perhaps more likely to escalate commitment, because they assumed personal
control over the situation and were optimistic about future gains (Lerner &
Keltner, 2000; Coleman, 2010).
The finding that students are no more likely to commit the sunk-cost
fallacy for an exam situation than for an interpersonal situation calls into
question how college students view emotions in romantic relationships and
friendships. Research by Labouvie-Vief et. al. (1989) and Carstensen and
Turk-Charles (1994) suggests that older adults are better at understanding
emotion states and are better at controlling emotions than are younger adults
such as college students. Thus, age differences in the salience of emotion
might help to explain why emotion was not the resource that predicted the
greatest sunk-cost effect for the interpersonal situations. Perhaps college
students are better at mentally accounting for time than they are for emotions.
A future study should further isolate these variables to determine which is
more predictive of future commitment in interpersonal relationships.
Effort
Like the present study’s results about the effects of emotions on both
situation types, results about invested effort revealed that students were also as
likely to commit the sunk-cost fallacy for the individual and interpersonal
situations. Effort has been repeatedly stated as one of the predictors of the
sunk-cost effect (Arkes & Blumer, 1985), but only recently has effort been
28
isolated to determine its effects on the sunk-cost effect. Coleman (2010a)
found that participants in the effort condition were the worst at mentally
accounting for effort when compared to participants in the time and money
conditions. Coleman concluded that this was the case, because either people
have not practiced keeping track of expended effort or did not care to keep
track of it.
Because I did find that people committed the sunk-cost fallacy for
effort in the exam situation, this might suggest that participants were
motivated to expend additional effort to do better on the exam. The theory of
learned industriousness (Eisenberger, 1992) may help to explain why a prior
investment of effort could lead to further investments of effort in an exam
situation. Learned industriousness theory suggests that rewarding the
completion of a difficult task may make the difficult task seem less aversive;
thus, people may be more likely to invest high effort. When this theory is
applied to the exam situation, it might seem plausible to imagine that if
students imagined that future invested effort would be rewarded by receiving
the score that they needed on the exam for graduate school admittance, then
they were perhaps more motivated to continue investing effort to see the
reward come to fruition.
Self-Concept
Similar to the results of both emotion and effort, results about the
effects of self-concept on future commitment revealed that participants were
just as likely to escalate commitment in the exam situation as they were to do
29
the same in the interpersonal situations. Put even more simply, the way
participants viewed themselves did not change their commitment behavior in
either situation type.
I predicted that cognitive dissonance theory (Festinger, 1957) would
account for escalation of commitment for the self-concept resource, or that
once people’s opinions of themselves came into conflict with a particular
choice (for example, viewing themselves as a caring friend but then faced
with the option of ending the friendship), people might be more likely to
remain committed to a particular course of action to avoid viewing themselves
as a bad friend for example. The results of the present study suggest that
dissonance may have occurred, leading participants to escalate commitment in
order to reduce a dissonant view of themselves. People may be likely to
escalate commitment in the presence of dissonance whether the situation is
individual such as the exam or interpersonal if their view of themselves is
consistent across situations.
Personality
Results did not support what others have found about the relationship
between personality and sunk-cost: conscientious and neurotic individuals are
perhaps more likely to commit the sunk-cost fallacy (Moon, 2001; Moon et
al., 2003). There are a couple explanations for the present study’s findings.
The personality inventory that I used was different from the one that Moon
and colleagues used. I did not split neuroticism into its constructs, therefore,
maybe I was unable to observe the effects that these specific constructs might
30
have had on sunk-cost behavior. Alternatively, our finding that personality
did not influence sunk-cost behavior may suggest that the situation is a better
predictor of behavior than personality. This explanation can be supported by
research that shows that personality is not consistent from situation to
situation and that there are low correlations between personality and behavior
(Kenrick & Funder, 1991; Mischel, 1968).
Sex
Although we did not manipulate sex in our study, the finding that men
had higher sunk-cost fallacy scores in the exam situation and women had
higher sunk-cost fallacy scores in the friend situation merits some discussion.
Unfortunately, it is not possible to know how men felt about the exam
situation, because I did not ask them. It could be possible that men became
angry at the thought of not doing well on the exam the first time around.
Consistent with research on emotion and escalation of commitment, if men
were emotionally aroused to the point of anger about the exam, then they
might have been more likely to commit the sunk-cost fallacy, because they
would have assumed personal control over the situation.
I also do not know why women committed the sunk-cost fallacy more
than men did for the friend situation, although research can provide an
explanation. Women are likely to engage in long-term care giving with a
friend, while men are more likely to engage in short-term heroic acts and be
done (Eagly, 1987). Similarly, Gabriel and Gardner (1999) found that men
spend more effort with groups while women spend more effort with individual
31
friends, similar to the situation in the present study’s vignette. Based on this
research, women are more likely than men to have the desire to escalate
commitment to an individual friend, similar to situation in the present study.
Limitations and Future Directions
There are several limitations in the present study. Perhaps one of the
biggest limitations is an inability to explain why participants committed the
sunk-cost fallacy for each level of resource and situation. While I thought that
personality would explain sunk-cost behavior, I found that it did not. Maybe
if I had included a more detailed personality inventory that split neuroticism
into its constructs of anxiety and depression, I might have found results more
consistent with previous research that linked neuroticism to the sunk-cost
effect (Moon, 2001). As the discussion of each resource may have revealed,
there might be several other explanations for why participants chose to
escalate commitment such as attribution theory, learned industriousness and
cognitive dissonance. I might have learned more about participants’ motives
behind their decisions had I asked them to explain why they made the decision
they made or by including, for example, an inventory that measured
participants’ anger and fear to enable us to determine whether those emotions
indeed influenced the sunk-cost effect.
Another limitation of the present study was the use of only one
scenario for the individual situation. While one situation helped me to observe
differences between both situation types, it might have been more revealing if
I had included another situation—perhaps about another topic that is relevant
32
to college students. Another individual situation may have enabled us to better
generalize our findings. Perhaps there are differences within individual
situations that might explain how these situations would differ with
interpersonal situations.
The present study opens up many different possibilities for future
research. Researchers could further investigate the role that self-concept plays
in the sunk-cost effect. They can also conduct studies to tease apart the role
that emotions and time play in people committing the sunk-cost fallacy.
The present study took on the enormous task of addressing many
limitations that have existed in previous sunk-cost research. The present study
adds to the literature by showing that the situation type does lead people to
commit future resources differently. In particular, college students may be
more likely to remain committed to interpersonal relationships even when
things are going sour because of time and money already spent building the
relationship; however, they are more likely to break off commitment to an
exam if they’d already invested a large amount of time preparing. The present
study has implications for intervention research that may want to target
particular resources that predict the escalation of commitment in a population
of college students.
33
References
A r k e s , H .R . , & Blu m e r, C . ( 1 98 5 ) . T h e ps yc h o l o g y o f
s unk cost . Organiz ati onal Behavi or and Human
Deci si on Process es , 35, 124-140.
A r k e s , H . R . , & Ay t o n , P. ( 1 9 9 9 ) . T h e s u n k - c o s t a n d
concorde effects: are humans less rational than
lower animals? Psychological Bulletin, 125, 59-1
600.
B o w e r , G. H . ( 1 9 8 1 ) . M o o d a n d m e m o r y. A m e r i c a n
Psychologist, 36, 129-148.
Br o c k n e r, J . (1 9 92 ) . T h e e s c a l at io n o f c om mi tm e nt to a
failing course of action. Academy of Management
Review, 17, 39-61.
C arst en s en , L. L. , & Tu rk -C h arl es , S. (1 99 4 ).
Psychology and Aging, 9, 259-264.
Coleman, M.D. (2009). Sunk cost and commitment to
dat es arranged onl i ne. C urrent Ps ychol ogy, 28,
45-54.
Coleman, M.D. (2010a). Sunk cost and commitment to
m edi cal t reatm ent. Cur rent Ps ychology, 29, 121134.
Coleman, M.D. (2010b). Sunk cost, emotion, and
com mi tm ent t o educat ion. Cur rent Ps ychol ogy, 29,
346-356
D i e n e r, E . , S u h , E. M . , Lu c a s , R . E. , & S mi th , H. L.
(1999). Subjective well-being: three decades of
progress. Psychological Bulletin, 125, 276-302.
E a g l y, A . H . ( 1 9 8 7 ) . S e x d i f f e r e n c e s i n s o c i a l b e h a v i o r :
a s o ci a l ro l e i n t er p re t a t i o n . H i l l s d a l e , N J :
Erlbaum.
E i s e nb e rge r, R. ( 19 9 2 ) . Le a r n e d i n du s t ri ou s n es s .
Psychological Review, 99, 248-267.
E l l i o t , A . J . , & D e v i n e , P. G. ( 1 9 9 4 ) . O n t h e
motivational nature of cognitive dissonance:
as psychological discomfort. Journal of
34
Personality and Social Psychology, 67, 382394.
Fe s t i n ge r, L. ( 1 9 57 ) . A t h eo r y o f cog n it i v e d is so n an c e .
Stanford, CA: Stanford University Press.
Frisch, D. (1993). Reasons for framing effects.
Organi zati onal Behavi or and Human Deci si on
Proces s es , 54, 399–429.
G a b r i e l , S . , & G a r d n e r , W. L . ( 1 9 9 9 ) . A r e t h e r e " h i s "
and "her" types of interdependence? The
implications of gender differences in collective
and relational interdependence for affect,
b e h a v io r, an d
cognition. Journal of Personality
and Social Psychology, 75, 642-655.
G o o d f r i e n d , W. , & A g n e w , C . R . ( 2 0 0 8 ) . S u n k e n c o s t s
and desired plans: examining different t ypes of
investments in close relationships. Personality
and Social Psychology Bulletin, 34, 1639–1652.
J o h n , O . P. , N a u m a n n , L . P. , & S o t o , C . J . ( 2 0 0 8 ) .
Paradigm shift to the integrative big-five trait
t a x o n o m y : h i s t o r y, m e a s u re m e n t , a n d c o n c e p t u a l
i s s u e s . N e w Yo r k : G u i l f o r d P r e s s .
K e n r i c k , D . T. , & F u n d e r , D . C . ( 1 9 9 1 ) . T h e p e r s o n
situation debate: do personality traits really
exist? . Personality: Contemporary Theory and
Res earch (pp. 150-174). Chi cago: Nel s on Hall .
K l a c z y n s k i , P. ( 2 0 0 1 ) . F r a m i n g e f f e c t s o n a d o l e s c e n t
task representations, anal ytic and heuristic
processing, and decision making: implications for
the normative/descriptive gap. Applied
Developmental Psychology, 22, 289–309.
L a b o u v i e - Vi e f , G. H . , D e Vo e , M . , & B u l k a , D . ( 1 9 8 9 ) .
Speaking about feelings: conceptions of emotion
across the life span. Psychology and aging, 4,
425-437.
L a b o u v i e - Vi e f , G. H . ( 1 9 9 2 ) . A n e o - P i a g e t i a n
perspective on adult cognitive development. In
R.J. Sternberg & C.A. Berg (Eds.), Intellectual
d e v e l o p m e n t ( p p . 5 2 - 8 6 ) . N e w Yo r k : C a m b r i d g e
University Press.
35
L e r n e r, J . S . , & K e l t n e r, D . ( 2 0 0 0 ) . B e yo n d v a l e n c e :
toward a model of emotion-specific influences on
judgement and choice. Cognition & Emotion, 14,
473-493.
M a r s h , H . W. , & S h a v e l s o n , R . J . ( 1 9 8 5 ) . S e l f - c o n c e p t :
its multifaceted, hierarchical structure.
Educational Psychologist, 20, 107-125.
M e t a l s k y, G. I . , A b r a m s o n , L . Y. , S e l i g m a n , M . E . ,
Semmel, A.J., & Peterson, C. (1982).
Attributional st yles and life events in the
classroom: vulnerability and invulnerability to
depressive mood reactions. Journal of Personality
and Social Psychology, 43, 612-617.
M i s c h e l , W. ( 1 9 6 8 ) . P e r s o n a l i t y a n d a s s e s s m e n t . N e w
Yo r k : W i l e y.
M o o n , H . W. ( 2 0 0 1 ) . T h e t w o f a c e s o f
conscientiousness: duty and achievement striving
in escalation of commitment dilemmas. Journal of
Applied Psychology, 86, 533-540.
M o o n , H . W. , H o l l e n b e c k , J . R . , H u m p h r e y, S . E . , &
Maue, B. (2003). The tripartite model of
neuroticism and the suppression of depression and
anxiety within an escalation of commitment
dilemma. Journal of Personality, 71, 347-368.
Phillips, O.R., Battalio, R.C., & Kogut, C.A. (1991).
Sunk and opportunity costs in valuation and
b i dd i n g. S o ut h er n Eco no mi c Jo u rn al , 5 8 , 11 2
128.
R a m m s t e d t , B . , & J o h n , O . P. ( 2 0 0 7 ) . M e a s u r i n g
personality in one minute or less: a 10-item short
version of the big five inventory in english and
germ an. Jour nal of Res earch i n Pers onal it y, 41,
203-212.
Rusbult, C.E. (1980a). Commitment and satisfaction in
romantic associations: a test of the investment
model. Journal of Experimental Social
Psychology, 16, 172-186.
Rusbult, C.E. (1980b). Satisfaction and commitment in
f r i e n d s h i p s . R e p re s e n t a t i v e R e s ea rc h i n S o ci a l
P s y c h o l o g y , 11 , 9 6 - 1 0 5 .
36
Rusbult, C.E. (1983). A longitudinal test of the
investment model: the development (and
deterioration) of satisfaction and commitment in
heterosexual involvements. Journal of Personality
and Social Psychology, 45, 172.
S h a v el so n , R .J ., H u b n e r, J .J ., & Sta n t o n, J .C . (1 9 76 ).
Self-concept: validation of construct
i nt erpret at ions. Revi ew of Educati onal Res earch,
406, 407-441.
S innot t, J . (1989). Ever yday probl em s ol ving: T heor y
a n d a p p l i c a t i o n s . N e w Yo r k : P r a e g e r .
Soman, D. (2001). The mental accounting of sunk
time costs: why time is not like money in
romantic associations: a test of the investment
model. Journal of Behavioral Decision Making,
14, 169–185.
Stanovich, K. (1999). Who is rational: studies of
i ndi vidual di ff erences i n reas oning. M ahwah, NJ :
Lawrence Erlbaum Associates, Inc.
S t a w, B . M . ( 1 9 7 6 ) . K n e e d e e p i n t h e b i g m u d d y: a
study of escalating commitment to a chosen
cours e of acti on. Organi zati onal Behavi or and
Human Performance, 16, 405-433.
S t a w, B . M . ( 1 9 8 1 ) . T h e e s c a l a t i o n o f c o m m i t m e n t t o a
failing course of action. Academy of Management
Review, 6, 577-587.
S t a w, B . M . , & H o a n g , H . ( 1 9 9 5 ) . S u n k c o s t s i n t h e
NBA: why draft order affects playi ng time and
survival in professional basketball.
Administrative Science Quarterly, 40, 474-494.
S t r o u g h , J . M . , M e h t a , C . M . , M c F a l l , J . P. , & S c h u l l e r ,
K.L. (2008). Are older adults less subject to the
sunk-cost fallacy than younger adults?
Psychological Science, 19, 650-652.
T v e r s k y, A . , & K a h n e m a n , D . ( 1 9 8 1 ) . T h e f r a m i n g o f
decisions and the psychology of choice. Science,
2 11 , 4 5 3 - 4 5 8 .
37
W h y t e , G. ( 1 9 8 6 ) . E s c a l a t i n g c o m m i t m e n t t o a
of action: a reinterpretation. Academy of
M a n a g e m e n t R e v i e w , 11 , 3 11 – 3 2 1 .
course
38
Figure
39
Capstone Summary
Background
Stop what you’re doing right now and recall a time you waited in a
long line at the movie theater to watch a movie that you begrudgingly decided
to give a try. Thirty minutes later you took your seat, watched the opening
credits, and realized that you should have watched something else. Did you
leave immediately or did you keep watching?
Now reflect on another situation: You were friends with someone for
two years and he or she abruptly stopped calling you, responding to your
messages, or being supportive of your future goals or interests. Did you break
off the friendship immediately, wait a little while longer to see if the
friendship get better, or did you stay committed? Did you have a romantic
relationship that seemed to be going bad? Did you stay? Did you leave?
Why people continue to commit further to an apparent failing course
of action is nothing new. Psychologists and economists refer to this
phenomenon as the “sunk-cost fallacy,” or “sunk-cost effect,” because the
prior investment that has already been made is “sunk” and it should have no
influence on future decisions (Arkes & Blumer, 1985). The reality is: People
do not like to waste what has already been invested, so they continue to invest
even more, particularly money, time, and effort (Arkes & Blumer, 1985).
Many psychologists and economists call this tendency to further commit an
escalation of commitment. Escalation of commitment and the sunk-cost
fallacy are related; when people perceive that they have invested too much to
quit” (Teger, 1980), they often tend to invest further—or escalate
40
commitment.
While a lot of fascinating research has been done on behavioral sunk
costs, unfortunately there are many limitations. For one, many articles discuss
the commonly cited three resources that predict the sunk-cost effect: money,
time, and effort, yet most of these studies have only focused on the effect that
money has on sunk-cost behavior. What about time and effort?
Within recent years, some researchers have begun to explore the
effects that time and effort have on the sunk-cost fallacy. Soman (2001) found
that money, but not time, predicted the sunk-cost effect, because people
cannot mentally account for time in the way that they account for money. To
extend Soman’s research, Coleman (2010) investigated the individual effects
of money, time, and effort. Consistent with Soman’s research, Coleman found
that money did predict the sunk-cost effect, but time and effort did not.
Coleman explained that people are not mentally good at accounting for effort
either.
Although Soman and Coleman have paved the way for future studies
to investigate the way that individual resources influence the sunk-cost
fallacy, there is still another big limitation in the field of sunk-cost to address:
There are opposing findings about sunk-cost effects in the literature about
sunk-costs in interpersonal relationships. For example, although Arkes and
Blumer have found that money predicts the sunk-cost fallacy in an individual
situation (a situation where one makes a decision by himself, such as
continuing to watch a boring movie), research by Goodfriend and Agnew
41
(2008) showed that money did not predict the sunk-cost effect in interpersonal
relationships, because people don’t tie happiness to money.
Methodology
Based on such a discrepancy in findings between the two situation
types, the present study investigated what would happen if both situation types
were manipulated in the same study. In particular, the initial research question
was: Whether the resources that people invest in an individual situation, for
example, would predict the sunk-cost effect, but that these same resources
would not predict the sunk-cost effect in an interpersonal situation? Several
resources that people invest into either situation type were selected after
research into both literatures on the sunk-cost effect. Money, time, and effort
were automatically selected, and then emotion and self-concept were chosen.
Emotion and self-concept are listed as resources that predict the sunk-cost
fallacy in the literature about sunk costs in interpersonal relationships.
However, emotion was nearly absent from the literature on sunk costs in
individual situations while self-concept was never mentioned.
A population of college students was selected, because research
indicates that this age group is more likely than older adults to commit the
sunk-cost fallacy, perhaps because younger adults focus more on losses than
gains (Strough et al., 2008). In addition, the present study wanted to
investigate why college students remain in relationships (friendships and
romantic relationships) even when things start to go bad.
Two hundred seven students at Syracuse University were recruited to
42
complete the study. Each student came in and completed a packet of questions
that pertained to one of the resources that was manipulated (money, time,
emotion, effort, and self-concept). Each packet had three different types of
situations (an exam situation, romantic relationship, and friend situation)
pertaining to the particular resource. For example, a student may have
received an exam situation for money that read, “You spent $30 preparing for
a standardized exam. After taking the exam, you received a score lower than
what’s required to be admitted to your program of choice. Considering all the
factors involved, how many more times will you take the exam?” The exam
situation represented the individual situation type, and this particular type of
situation was chosen because it is one that college students have had
experience with.
Each of the three situations in the packet had a parallel situation but
with another level of investment. For example, one vignette may have asked
students about spending $30 (which was regarded as the low investment), but
students were also asked the same question, only with a different level of
investment: $1,500 (the high investment). To compute students’ sunk-cost
fallacy scores, their selections for the high and low investment conditions
were compared. If students chose to take the exam more times because of a
large investment paid, then they committed the sunk-cost fallacy.
After students completed the packet, they completed a brief
personality inventory, because previous research suggests that
conscientiousness and neuroticism can predict the sunk-cost effect (Moon
43
2001; Moon et al. 2003). Students then filled out a demographics
questionnaire that asked them questions such as their age, sex, and they were
also asked to rate the resources in order of importance for each situation type.
Results and Discussion
The type of situation and the type of resource did interact with each
other to exert an influence on sunk-cost fallacy scores. Students were more
likely to spend more time in a friendship and romantic relationship if they had
previously invested time, but they were less likely to continue taking an exam
if they had previously invested a large amount of time. Students were slightly
more likely to remain committed to a romantic relationship if money had been
invested than they were to continue taking an exam if they had invested
money. While students committed the sunk-cost fallacy for emotion, effort,
and self-concept, the difference between situation types were not statistically
significant.
The reasons behind the present study’s findings are explained in depth
in the Capstone paper; however, the major finding in this study is that invested
time overwhelmingly predicts future commitment in college relationships
(friendships and romantic relationships). This finding has implications for
intervention studies that might want to investigate ways to alter sunk-cost
behavior, at least among college students. The present study investigated the
individual role of different types of resources in different types of situations to
give researchers more insight on the type of resources that should be targeted
in an intervention. Based on the results of the present study, teaching students
44
that previously invested time is irretrievable—and thus should not have an
impact on future decisions—miight be a good start.